Immune and stromal scoring system associated with tumor microenvironment and prognosis: a gene-based multi-cancer analysis

Zihang Zeng,Yangyi Li,Qiuji Wu,Jianguo Zhang,Jiali Li,Xingyu Liu,Zhengrong Huang,Conghua Xie,Jiarui Chen,Yan Gong

doi:10.1186/s12967-021-03002-1

Zihang Zeng, Yangyi Li + Show 8 more

Open Access

https://doi.org/10.1186/s12967-021-03002-1

Copy DOI

Abstract

BackgroundTumor microenvironment (TME) is associated with tumor progression and prognosis. Previous studies provided tools to estimate immune and stromal cell infiltration in TME. However, there is still a lack of single index to reflect both immune and stromal status associated with prognosis and immunotherapy responses.MethodsA novel immune and stromal scoring system named ISTMEscore was developed. A total of 15 datasets were used to train and validate this system, containing 2965 samples from lung adenocarcinoma, skin cutaneous melanoma and head and neck squamous cell carcinoma.ResultsThe patients with high immune and low stromal scores (HL) were associated with low ratio of T cell co-inhibitory/stimulatory molecules and low levels of angiogenesis markers, while the patients with low immune and high stromal scores (LH) had the opposite characteristics. The HL patients had immune-centered networks, while the patients with low immune and low stromal scores (LL) had desert-like networks. Moreover, copy number alteration burden was decreased in the HL patients. For the clinical characteristics, our TME classification was an independent prognostic factor. In the 5 cohorts with immunotherapy, the LH patients were linked to the lowest response rate.ConclusionsISTMEscore system could reflect the TME status and predict the prognosis. Compared to previous TME scores, our ISTMEscore was superior in the prediction of prognosis and immunotherapy response.

Highlights

Tumor microenvironment (TME) is associated with tumor progression and prognosis
Through l2,1-norm Multitask learning (MTL), we identified the genes associated with TME-related low dimensional features (LDF)
To isolated the TMErelated signals from mixed tumor tissues, we first divided RNA-seq data of the training cohort (TCGA LUAD) [11] into 11 heterogeneous clusters based on negative matrix factorization (NMF) (Fig. 3A)

Summary

Introduction

Tumor microenvironment (TME) is associated with tumor progression and prognosis. Previous studies provided tools to estimate immune and stromal cell infiltration in TME. There is still a lack of single index to reflect both immune and stromal status associated with prognosis and immunotherapy responses. With the discovery of immune checkpoint molecules, immunotherapy becomes more promising strategy for cancer patients to elicit clinical responses durably [1]. A range of algorithms have been developed to estimate the immune and stromal cell infiltration including CIBERSORT, TIMER, ESTIMATE and MCPcounter [6,7,8,9]. These tools perform well in the estimation of TME cells, but not in the prediction of tumor prognosis and immunotherapy responses. There is still a lack of single index to reflect both immune and stromal activation signals associated with prognosis and immunotherapy responses

Methods

Results

Discussion

Conclusion